To say that University of Cambridge deployed disparate and diverse data storage and data center infrastructure would be a vast understatement.
In a 2013 IT review, data center manager Ian Tasker and his team discovered almost 200 servers across the University’s 120 departments. These installations ranged from single servers housed in closets to larger rooms containing 20 to 30 servers, but all contributing to a major drain on power and cooling resources, as well as creating one heck of a management headache.
“Our IT review is a periodic way for us to look at how we’re carrying out the responsibilities of IT, and how to best align IT with the work of the university. What we found was everything was fragmented and each department was doing their own thing when it came to provisioning, storage and management. Where we had a small number of racks, they weren’t powered or cooled efficiently, and that was our other mission: to reduce our carbon footprint as a university by about 34 percent by the year 2020,” Tasker says.
A £20 million (more than $30 million) data center overhaul and consolidation project was approved to address both of these challenges; standardizing and easing provisioning, storage and information management and increasing power and cooling efficiency, says Tasker.
Getting a Grip
With such a disparate and disjointed infrastructure, one of the major initial challenges was determining where computing resources were located as well as measuring their power consumption, says Tasker.
“One of our biggest challenges was understanding what was going on in our overall environment. Monitoring usage, looking at general power consumption, where and how resources, cooling and power are used, was very difficult because we didn’t have a single source of information if there was information available at all,” says Tasker.
Complexity was the major issue. Tasker says that with so many different deployments, IT needed to get control over all the assets to be able to monitor resource usage and then figure out how to best manage them to meet the needs of the university.
Scalability, Flexibility, Growth Potential
Another challenge Tasker faced was the issue of scalability. The data center consolidation needed to start small and then scale to encompass the entire university, but needed to be accomplished without disrupting day-to-day operations.
“We had to consider how to consolidate multiple sites, and we had to ensure flexibility, too. We needed a solution that would allow us a single-pane view of everything: a building management system (BMS) to automate power and cooling; energy management, data monitoring. And we needed the system to be dynamic so we could deliver adequate resources and services based on the changing needs of various departments, whether that was research, engineering, administration and the like,” says Tasker.
Planning for the Future
Initially, the data center will serve University Information Services (UIS), the High Performance Computing Service (HPCS) supporting the University’s research activities and the administrative needs of Cambridge Assessment’s three examination boards, according to UIS. Cambridge University Press also will share the data center.
The new data center facility includes four data halls, each designed to meet the different IT density requirements of its key stakeholders, according to UIS. There are currently 60 racks across three halls: Hall 1 accommodates the HPCS’s high density IT load of up to 900KW; Hall 2 provides 201KW for Cambridge Assessment’s needs, and Hall 3, 240KW for UIS’s servers, according to UIS.
Hall 4 remains deliberately unallocated, and has not been fully fitted out, UIS says. This will allow for 40 to 50 additional racks as future data center demand increases.
To monitor and manage the infrastructure, Tasker and his team chose Emerson Network Power’s Trellis platform. The solution addressed all the major issues facing Tasker and his team: consolidation, efficiency, single-pane monitoring and management, flexibility and scalability.
A Single-Pane Solution
“This will help us drastically improve our ability to do more detailed capacity management and rapidly address capacity issues. In the past, we haven’t been able to respond quickly when people get new products or need to install new solutions — there’s nowhere to put it because there hasn’t been adequate future capacity planning. Now, we can create and plan better capabilities for more research and development spaces,” says Tasker.
The new data center also goes a long way toward meeting the university’s goal of cutting its carbon footprint, according to UIS. University carbon emissions are expected to decline by at least 10 percent as a result of the data center consolidation, a significant improvement over current levels, UIS says.
UIS also predicts great improvements in power usage effectiveness (PUE), which is a measure of the ratio of power used by a data center to the power consumed by the actual IT equipment; a perfect PUE would be 1. Major IT players such as Google and Facebook have made significant investments in their data centers, and boast PUEs of 1.11 and 1.06 – 1.08, respectively. When it’s fully operational, Cambridge UIS predicts its new data center will deliver an overall PUE of 1.2, half of the January 2013 European average of 2.5.
The Data Center of the Future
The data center is currently undergoing final testing, and new technical staff are being recruited to operate the facility, according to UIS. Over the next few months a phased transition approach will see equipment relocated to the data center, with much of it planned to be operational by year end, UIS says.
The most important thing to remember when undertaking such a massive transformation is to view it as a journey, not just a destination, and to make sure that the vendors you choose are aligned not just with your organization’s current needs, but with long-term goals, Tasker says.
“You have to have a solid strategy in place and be able to articulate what you’re trying to achieve in the short- and long-term. Don’t underestimate how much effort and insight must go into planning and executing your strategy, and work with your vendors to push the limits of technology so you know you’re prepared for everything that lies ahead,” says Tasker.